Quality by Design in APIs

Published on: 

The adoption of quality by design in small-molecule drug development and manufacturing continues to evolve as the industry seeks ways to augment process understanding for APIs.

The science- and risk-based approach in quality by design (QbD) entails greater product and process understanding as a means to ensure product quality. These concepts are embodied in ICH guidelines Q8 (R2) Pharmaceutical Development, Q9 Quality Risk Management, Q10 Pharmaceutical Quality System, and most recently, Q11 Development and Manufacture of Drug Substances (Chemical Entities and Biotechnological/Biological Entities) (1-4), which offer a lifecycle approach to continual improvement to drug manufacturing.  

Traditional versus enhanced approaches
ICH Q11 focuses specifically on the development and manufacture of drug substances. It specifies that a company can follow “traditional” or “enhanced” approaches or a combination of both in developing a drug substance (4). In the traditional approach, set points and operating ranges for process parameters are defined, and the drug-substance control strategy is typically based on process reproducibility and testing to meet established acceptance criteria (4). In an enhanced approach, risk management and scientific knowledge are used more extensively to identify and understand process parameters and unit operations that affect critical quality attributes (CQAs) (4). The enhanced approach further includes the development of appropriate control strategies applicable over the lifecycle of the drug substance that may include the establishment of design space(s) (4). Manufacturing process development should include, at a minimum, according to ICH Q11:

  • Identifying potential CQAs associated with the drug substance so that those characteristics having an impact on drug-product quality can be studied and controlled

  • Defining an appropriate manufacturing process

  • Defining a control strategy to ensure process performance and drug-substance quality (4).

An enhanced approach to manufacturing process development would additionally include:


  • A systematic approach to evaluating, understanding, and refining the manufacturing process, including identifying--through prior knowledge, experimentation and risk assessment--the material attributes (e.g., of raw materials, starting materials, reagents, solvents, process aids, intermediates) and process parameters that can have an effect on drug-substance CQAs

  • Determining the functional relationships that link material attributes and process parameters to drug-substance CQAs (4).

The enhanced approach, in combination with quality risk management, is used to establish an appropriate control strategy. Those material attributes and process parameters important to drug-substance quality should be addressed by the control strategy. The risk assessment can include an assessment of manufacturing process capability, attribute detectability, and severity of impact as they relate to drug-substance quality (4). For example, when assessing the link between an impurity in a raw material or intermediate and drug-substance CQAs, the ability of the drug-substance manufacturing process to remove that impurity or its derivatives should be considered in the assessment (4). The risk related to impurities can usually be controlled by specifications for raw material/intermediates and/or robust purification capability in downstream steps. It is important to understand the formation, fate (whether the impurity reacts and changes its chemical structure), and purge (whether the impurity is removed via crystallization, extraction, etc.) as well as their relationship to the resulting impurities that end up in the drug substance as CQAs (4). The process should be evaluated to establish appropriate controls for impurities as they progress through multiple process operations (4).

Regulatory expectations
In March 2011, the European Medicines Agency and Food and Drug Administration launched a three-year pilot program for a parallel assessment by both agencies of certain quality and chemistry, manufacturing and control (CMC) sections of new drug applications submitted to FDA and marketing authorization applications (MAAs) submitted to EMA that are relevant to QbD, such as development, design space, and real-time release testing. The objective of the pilot, which is scheduled to end Mar. 31, 2014, is to ensure consistent implementation between the European Union and the United States of ICH guidelines Q8, Q9, Q10, and Q 11 and to facilitate sharing of regulatory decisions (5-7).

In August 2013, the agencies reported that the first EMA-FDA parallel assessment of QbD elements of an initial MAA was successfully finalized as well as some consultative advice procedures. In a question-and-answer format, EMA and FDA reported on their expectations as they relate to quality target product profiles (QTPPs), CQAs, classification of criticality, and application of QbD in analytical method development (7).

With respect to the QTPP, the agencies specified that they expect applicants to provide the QTPP, which describes prospectively the quality characteristics of a drug product that should be achieved to ensure the desired quality, safety, and efficacy of the drug product. With respect to CQAs, the agencies expect applicants to provide a list of CQAs for the drug substance, finished product, and excipients when relevant. This list should also include the acceptance limits for each CQA and a rationale for designating these properties as a CQA. Furthermore, there should be a discussion of how the drug substance and excipient CQAs relate to the finished product CQAs based on prior knowledge, risk assessment, or experimental data. The basis of the control strategy is to ensure that the drug substance and finished product CQAs are consistently within the specified limits (7).

Another issue was whether the agencies would accept a three-tier classification of criticality for process parameters. The agencies responded that ICH Q8 (R2) specifies that a critical process parameter is one whose variability has an impact on a CQA and needs to be monitored or controlled to ensure the process produces the desired quality. EMA and FDA cited a regulatory submission in which the applicant proposed an approach to risk assessment and determination of criticality that included a three-tier classification (“critical,” “key,” and “noncritical”) for quality attributes and process parameters. Using this classification, a “critical” factor was defined as a factor that led to failure during experimentation. A factor that had not led to failure within the range studied, but still may have an impact on product quality, was considered as a “key” factor. The agencies said that they do not support the use of the term “key process parameters (KPP)” because it is not ICH terminology and there is differing use of the term “key” among applicants. Although FDA and EMA said they are amenable to this terminology in the pharmaceutical development section to communicate development findings, they are not in describing the manufacturing process, process control, and control of critical steps and intermediates, where the description of all parameters that have an impact on a CQA should be classified as critical (7).

The agencies further specified that process manufacturing descriptions be comprehensive and describe process steps in a sequential manner, including batch size(s) and equipment type. The critical steps and points at which process controls, intermediate tests, or final product controls are conducted should be identified (7). Steps in the process should have the necessary detail in terms of appropriate process parameters along with their target values or ranges. The process parameters that are included in the manufacturing process description should not be restricted to the critical ones; all parameters that have been demonstrated during development as needing to be controlled or monitored during the process to ensure that the product is of the intended quality need to be described (7).

The agencies also commented on QbD as it relates to analytical methods using risk assessments and statistically designed experiments to define analytical target profiles (ATP) and method operational design ranges (MODR) for analytical methods (7). “There is currently no international consensus on the definition of ATP and MODR,” noted the agencies. “Until this is achieved, any application that includes such proposals will be evaluated on a case-by-case basis” (7). The agencies noted, however, that an ATP can be acceptable as a qualifier of the expected method performance by analogy to the QTPP as defined in ICH Q8 (R2), but the agencies would not consider analytical methods that have different principles (e.g., high-performance liquid chromatography  and near-infrared [NIR] spectroscopy) equivalent solely on the basis of conformance with the ATP. “An applicant should not switch between these two types of methods without appropriate regulatory submission and approval,” they said. The agencies also noted that similar principles and data requirements could apply for MODRs. For example, data to support an MODR could include: appropriately chosen experimental protocols to support the proposed operating ranges/conditions and demonstration of statistical confidence throughout the MODR. Issues for further reflection include the assessment of validation requirements as identified in ICH Q2 (R1) throughout the MODR and confirmation of system suitability across all areas of the MODR (7). The agencies further indicated that future assessment of the pilot program will include other lessons learned in areas such as design-space verification, the level of detail in submissions for design space and risk assessment, continuous process verification, and continuous manufacturing.

QbD at work
A review of recent literaure reveals some interesting applications of QbD in drug-substance development and manufacturing. For example, scientists at Bristol-Myers Squibb reported on a process-modeling method using a QbD approach in the development of the API ibipinabant, a cannabinoid receptor 1 antagonist being developed to treat obesity (8). In its development, the molecule had volume requirements of 6 kg for toxicology studies and formulation development, which later increased to 175 kg for late-stage clinical trials. The researchers used mechanistic kinetic modeling to understand and control undesired degradation of enantiomeric purity during API crystallization. They implemented a work flow, along with kinetic and thermodynamic process models, to support the underlying QbD approach and reported on the use of risk assessment, target quality specifications, operating conditions for scale-up, and plant control capabilities to develop a process design space (8). Subsequent analysis of process throughput and yield defined the target operating conditions and normal operating ranges for a specific pilot-plant implementation. Model predictions were verified from results obtained in the laboratory and at the pilot-plant scale (8). Future efforts were focused on increasing fundamental process knowledge, improving model confidence, and using a risk-based approach to re-evaluate the design space and select operating conditions for the future scale-up (8).

Scientists at Merck & Co. reported on their work in applying QbD to set up an improved control strategy for the final five steps in the production route of a legacy steroidal contraceptive, which has been produced for more than 20 years within its facilities (9). A generic ultra-high-performance liquid chromatography method was developed according to QbD principles to create a range of proven acceptance criteria for the assay and side-product determination for the final five steps in the production route of the API (9).

Scientists at Eli Lilly reported on a systematic approach consisting of a combination of first-principles modeling and experimentation for the scale-up from bench to pilot-plant scale to estimate the process performance at different scales and study the sensitivity of a process to operational parameters, such as heat-transfer driving force, solvent recycle, and removed fraction of volatiles (10). This approach was used to predict process outcomes at the laboratory and pilot-plant scale and to gain a better understanding of the process. The model was also used further to map the design space (10).

In other work, scientists reported on the application of latent variables-based modeling to a reaction process in a small-molecule synthesis based on continuous-flow hydrogenation (11). In another study, scientists reported on using a QbD approach for designing improved stability studies (12). Also, scientists at UCB and the Institut des Sciences Moléculaires de Marseille in France recently reported on the feasibility of using online NIR spectroscopy as a process analytical technology tool to monitor in real time the API and residual solvent content to control the seeding of an API crystallization process at industrial scale. A quantitative method was developed at laboratory scale using statistical design of experiments and multivariate data analysis (13).

1. ICH, Q8 (R2) Pharmaceutical Development (2009).
2. ICH, Q9 Quality Risk Management (2005).
3. ICH, Q10 Pharmaceutical Quality System (2008).
4. ICH, Q11 Development and Manufacture of Drug Substances (Chemical Entities and Biotechnological/Biological Entities)(2012).
5. FDA, “FDA, EMA Announce Pilot for Parallel Assessment of Quality by Design Applications,” Press Release, Mar. 16, 2011.
6. EMA, “European Medicines Agency and US Food and Drug Administration Announce Pilot Program for Parallel Assessment of Quality by Design Applications,” Press Release, Mar. 16, 2011.
7. EMA, “EMA-FDA Pilot Program for Parallel Assessment of Quality-by-Design Applications: Lessons Learnt and Q&A Resulting from the First Parallel Assessment,” Press Release, Aug. 20, 2013.
8. S.B. Brueggemeier et al., Org. Proc. Res. Dev. 16 (4) 567-576 (2012).
9. J. Musters et al., Org. Process Res. Dev.
17 (1) 87-96 (2013).
10. I. Figueroa, S. Vaidyaraman and S. Viswanath, Org. Process Res. Dev., online, DOI: 10.1021/op4001127, July 22, 2013.
11. Z. Shi, N. Zaborenkdo, and D.E. Reed,
J. Pharm. Innov. 8 (10) 1-10 (2013).
12. S.T. Colgan, J. Pharm. Innov. 7 (3-4)
205-213 (2012).
13. C. Schaefer et al., J. Pharm. Biomed. Anal., online, DOI.org/10.1016/j.jpba.2013.05.015. May 20, 2013. PT